I once intentionally crashed Carnegie Mellon University’s computer system

The Andrew system (as I have been told) tends to run at about 130% or its prime operating system. What this obscene statistic translates to is that it is typically pushing its resources to the point where the slightest thing could really mess it up.

About 7-10 years ago the ‘net’ in Pittsburgh went to absolute hell one afternoon across the city. An industrial backhoe had been digging without checking with the PUC somewhere around one of the surrounding counties. The backhoe had managed to take out one of the T-3 bundles that fed the Pittsburgh Supercomputing Center, which in turn (I’m given the ‘correctable’ impression) at the time fed most of the city’s universities and ISPs.

Service at Pitt and CMU went to absolute hell. The systems were running, but since their mere operation was so dependant upon the internet, they were more than a little flaky. A colleague and I were looking for something net related (a usenet archive) on one of the served Andrew systems. (Andrew had this strange method of email and news that allowed you to actually look at the USENet hierarchy as a series of folders with cryptically named files. I asked my colleague if we could do a recursive search for a directory. I was looking to see if there was a news group that had the name ‘foobarsomethingorother’ in it.

This was a very innocuous command, something to the effect of:

ls -alR /andrew/usenet/ | grep fly-fishing

There was a longer than average pause which we credited to the fact that the net had wreaked havoc on the system, but we figured the directories were local, so this should not have been a bad task. Suddenly the terminal started sending us strange errors:

Vice 2 going down
Vice 3 going down
…
Vice 23 going down

It was explained to me that those were the mounted user system partition mirrors which kept the files for all users and other necessary data. After those messages, his account was no longer accessible and his directory was ‘temporary unaccessible.’ We looked at each other. This thing happened on occasion and it tended to mean something sorta crashed and needed to re’something’ and come back up. The process took about 4 minutes to fix itself.

After 4 minutes… Vice 2 is up….etc

“Did we do that”, I asked him
“Dunno.” he replied.

I hit the up arrow to cycle to the last search command and intentionally hit enter.

“What are you doing?” he asked.
“Finding out if it was coincidence or us…”

Vice 2 going down
Vice 3 going down
…
Vice 23 going down

And then 4 minutes later it was back up again.

“So what was the point of that?”
“Now I can email their management and tell them to investigate a nasty bug in their system”

——-

A few years later I was applying for a software engineering position at Pitt with an emphasis in networking and security. During the interview the following exchange occurred:

“Are you familiar with large scale networking dynamics and fault tolerance management.”
“I crashed CMU once”
“Intentionally ?!?!?!”
“Not the first time.”
“Excuse me”
“I had to go back and find out if it was me, so I could send the issue to someone for diagnosis.”

And 3 days later I was hired. I was notified by the manager that that comment was the one that put me ahead of the other guy who had my level of experience and broke the tie.

Summary: Hacker… Was as a kid, discovered that mucking with a system can be dangerous and it’s better to be helpful to them than an outsider that they can come after. By college age, I was trying to be knowledgeable without being intrusive. Anything that went wrong was not as a result of me trying to do something that one was not supposed to do.

« »